Spatio-Temporal Analysis of Wikipedia Metadata and the STiki Anti-Vandalism Tool

Track: Posters

Authors: Andrew West, Sampath Kannan and Insup Lee (University of Pennsylvania, USA)

Poster: POSTER

Abstract
The bulk of Wikipedia anti-vandalism tools require natural language processing over the article or diff text. However, our prior work demonstrated the feasibility of using spatio-temporal properties to locate malicious edits. STiki is a real-time, on-Wikipedia tool leveraging this technique. The associated poster reviews STiki’s methodology and performance. We find competing anti-vandalism tools inhibit maximal performance. However, the tool proves particularly adept at mitigating long-term embedded vandalism. Further, its robust and language-independent nature make it well-suited for use in less-patrolled Wiki installations.